[deliverable/linux.git] / net / ipv4 / inet_timewait_sock.c

/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		Generic TIME_WAIT sockets functions
 *
 *		From code orinally in TCP
 */

#include <linux/kernel.h>
#include <linux/kmemcheck.h>
#include <linux/slab.h>
#include <linux/module.h>
#include <net/inet_hashtables.h>
#include <net/inet_timewait_sock.h>
#include <net/ip.h>


/**
 *	inet_twsk_unhash - unhash a timewait socket from established hash
 *	@tw: timewait socket
 *
 *	unhash a timewait socket from established hash, if hashed.
 *	ehash lock must be held by caller.
 *	Returns 1 if caller should call inet_twsk_put() after lock release.
 */
int inet_twsk_unhash(struct inet_timewait_sock *tw)
{
	if (hlist_nulls_unhashed(&tw->tw_node))
		return 0;

	hlist_nulls_del_rcu(&tw->tw_node);
	sk_nulls_node_init(&tw->tw_node);
	/*
	 * We cannot call inet_twsk_put() ourself under lock,
	 * caller must call it for us.
	 */
	return 1;
}

/**
 *	inet_twsk_bind_unhash - unhash a timewait socket from bind hash
 *	@tw: timewait socket
 *	@hashinfo: hashinfo pointer
 *
 *	unhash a timewait socket from bind hash, if hashed.
 *	bind hash lock must be held by caller.
 *	Returns 1 if caller should call inet_twsk_put() after lock release.
 */
int inet_twsk_bind_unhash(struct inet_timewait_sock *tw,
			  struct inet_hashinfo *hashinfo)
{
	struct inet_bind_bucket *tb = tw->tw_tb;

	if (!tb)
		return 0;

	__hlist_del(&tw->tw_bind_node);
	tw->tw_tb = NULL;
	inet_bind_bucket_destroy(hashinfo->bind_bucket_cachep, tb);
	/*
	 * We cannot call inet_twsk_put() ourself under lock,
	 * caller must call it for us.
	 */
	return 1;
}

/* Must be called with locally disabled BHs. */
static void inet_twsk_kill(struct inet_timewait_sock *tw)
{
	struct inet_hashinfo *hashinfo = tw->tw_dr->hashinfo;
	struct inet_bind_hashbucket *bhead;
	int refcnt;
	/* Unlink from established hashes. */
	spinlock_t *lock = inet_ehash_lockp(hashinfo, tw->tw_hash);

	spin_lock(lock);
	refcnt = inet_twsk_unhash(tw);
	spin_unlock(lock);

	/* Disassociate with bind bucket. */
	bhead = &hashinfo->bhash[inet_bhashfn(twsk_net(tw), tw->tw_num,
			hashinfo->bhash_size)];

	spin_lock(&bhead->lock);
	refcnt += inet_twsk_bind_unhash(tw, hashinfo);
	spin_unlock(&bhead->lock);

	BUG_ON(refcnt >= atomic_read(&tw->tw_refcnt));
	atomic_sub(refcnt, &tw->tw_refcnt);
	atomic_dec(&tw->tw_dr->tw_count);
	inet_twsk_put(tw);
}

void inet_twsk_free(struct inet_timewait_sock *tw)
{
	struct module *owner = tw->tw_prot->owner;
	twsk_destructor((struct sock *)tw);
#ifdef SOCK_REFCNT_DEBUG
	pr_debug("%s timewait_sock %p released\n", tw->tw_prot->name, tw);
#endif
	kmem_cache_free(tw->tw_prot->twsk_prot->twsk_slab, tw);
	module_put(owner);
}

void inet_twsk_put(struct inet_timewait_sock *tw)
{
	if (atomic_dec_and_test(&tw->tw_refcnt))
		inet_twsk_free(tw);
}
EXPORT_SYMBOL_GPL(inet_twsk_put);

static void inet_twsk_add_node_rcu(struct inet_timewait_sock *tw,
				   struct hlist_nulls_head *list)
{
	hlist_nulls_add_head_rcu(&tw->tw_node, list);
}

static void inet_twsk_add_bind_node(struct inet_timewait_sock *tw,
				    struct hlist_head *list)
{
	hlist_add_head(&tw->tw_bind_node, list);
}

/*
 * Enter the time wait state. This is called with locally disabled BH.
 * Essentially we whip up a timewait bucket, copy the relevant info into it
 * from the SK, and mess with hash chains and list linkage.
 */
void __inet_twsk_hashdance(struct inet_timewait_sock *tw, struct sock *sk,
			   struct inet_hashinfo *hashinfo)
{
	const struct inet_sock *inet = inet_sk(sk);
	const struct inet_connection_sock *icsk = inet_csk(sk);
	struct inet_ehash_bucket *ehead = inet_ehash_bucket(hashinfo, sk->sk_hash);
	spinlock_t *lock = inet_ehash_lockp(hashinfo, sk->sk_hash);
	struct inet_bind_hashbucket *bhead;
	/* Step 1: Put TW into bind hash. Original socket stays there too.
	   Note, that any socket with inet->num != 0 MUST be bound in
	   binding cache, even if it is closed.
	 */
	bhead = &hashinfo->bhash[inet_bhashfn(twsk_net(tw), inet->inet_num,
			hashinfo->bhash_size)];
	spin_lock(&bhead->lock);
	tw->tw_tb = icsk->icsk_bind_hash;
	WARN_ON(!icsk->icsk_bind_hash);
	inet_twsk_add_bind_node(tw, &tw->tw_tb->owners);
	spin_unlock(&bhead->lock);

	spin_lock(lock);

	/*
	 * Step 2: Hash TW into tcp ehash chain.
	 * Notes :
	 * - tw_refcnt is set to 3 because :
	 * - We have one reference from bhash chain.
	 * - We have one reference from ehash chain.
	 * We can use atomic_set() because prior spin_lock()/spin_unlock()
	 * committed into memory all tw fields.
	 */
	atomic_set(&tw->tw_refcnt, 1 + 1 + 1);
	inet_twsk_add_node_rcu(tw, &ehead->chain);

	/* Step 3: Remove SK from hash chain */
	if (__sk_nulls_del_node_init_rcu(sk))
		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);

	spin_unlock(lock);
}
EXPORT_SYMBOL_GPL(__inet_twsk_hashdance);

static void tw_timer_handler(unsigned long data)
{
	struct inet_timewait_sock *tw = (struct inet_timewait_sock *)data;

	if (tw->tw_kill)
		NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITKILLED);
	else
		NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITED);
	inet_twsk_kill(tw);
}

struct inet_timewait_sock *inet_twsk_alloc(const struct sock *sk,
					   struct inet_timewait_death_row *dr,
					   const int state)
{
	struct inet_timewait_sock *tw;

	if (atomic_read(&dr->tw_count) >= dr->sysctl_max_tw_buckets)
		return NULL;

	tw = kmem_cache_alloc(sk->sk_prot_creator->twsk_prot->twsk_slab,
			      GFP_ATOMIC);
	if (tw) {
		const struct inet_sock *inet = inet_sk(sk);

		kmemcheck_annotate_bitfield(tw, flags);

		tw->tw_dr	    = dr;
		/* Give us an identity. */
		tw->tw_daddr	    = inet->inet_daddr;
		tw->tw_rcv_saddr    = inet->inet_rcv_saddr;
		tw->tw_bound_dev_if = sk->sk_bound_dev_if;
		tw->tw_tos	    = inet->tos;
		tw->tw_num	    = inet->inet_num;
		tw->tw_state	    = TCP_TIME_WAIT;
		tw->tw_substate	    = state;
		tw->tw_sport	    = inet->inet_sport;
		tw->tw_dport	    = inet->inet_dport;
		tw->tw_family	    = sk->sk_family;
		tw->tw_reuse	    = sk->sk_reuse;
		tw->tw_hash	    = sk->sk_hash;
		tw->tw_ipv6only	    = 0;
		tw->tw_transparent  = inet->transparent;
		tw->tw_prot	    = sk->sk_prot_creator;
		atomic64_set(&tw->tw_cookie, atomic64_read(&sk->sk_cookie));
		twsk_net_set(tw, sock_net(sk));
		setup_timer(&tw->tw_timer, tw_timer_handler, (unsigned long)tw);
		/*
		 * Because we use RCU lookups, we should not set tw_refcnt
		 * to a non null value before everything is setup for this
		 * timewait socket.
		 */
		atomic_set(&tw->tw_refcnt, 0);

		__module_get(tw->tw_prot->owner);
	}

	return tw;
}
EXPORT_SYMBOL_GPL(inet_twsk_alloc);

/* These are always called from BH context.  See callers in
 * tcp_input.c to verify this.
 */

/* This is for handling early-kills of TIME_WAIT sockets. */
void inet_twsk_deschedule(struct inet_timewait_sock *tw)
{
	if (del_timer_sync(&tw->tw_timer))
		inet_twsk_kill(tw);
}
EXPORT_SYMBOL(inet_twsk_deschedule);

void inet_twsk_schedule(struct inet_timewait_sock *tw, const int timeo)
{
	/* timeout := RTO * 3.5
	 *
	 * 3.5 = 1+2+0.5 to wait for two retransmits.
	 *
	 * RATIONALE: if FIN arrived and we entered TIME-WAIT state,
	 * our ACK acking that FIN can be lost. If N subsequent retransmitted
	 * FINs (or previous seqments) are lost (probability of such event
	 * is p^(N+1), where p is probability to lose single packet and
	 * time to detect the loss is about RTO*(2^N - 1) with exponential
	 * backoff). Normal timewait length is calculated so, that we
	 * waited at least for one retransmitted FIN (maximal RTO is 120sec).
	 * [ BTW Linux. following BSD, violates this requirement waiting
	 *   only for 60sec, we should wait at least for 240 secs.
	 *   Well, 240 consumes too much of resources 8)
	 * ]
	 * This interval is not reduced to catch old duplicate and
	 * responces to our wandering segments living for two MSLs.
	 * However, if we use PAWS to detect
	 * old duplicates, we can reduce the interval to bounds required
	 * by RTO, rather than MSL. So, if peer understands PAWS, we
	 * kill tw bucket after 3.5*RTO (it is important that this number
	 * is greater than TS tick!) and detect old duplicates with help
	 * of PAWS.
	 */

	tw->tw_kill = timeo <= 4*HZ;
	if (!mod_timer_pinned(&tw->tw_timer, jiffies + timeo)) {
		atomic_inc(&tw->tw_refcnt);
		atomic_inc(&tw->tw_dr->tw_count);
	}
}
EXPORT_SYMBOL_GPL(inet_twsk_schedule);

void inet_twsk_purge(struct inet_hashinfo *hashinfo,
		     struct inet_timewait_death_row *twdr, int family)
{
	struct inet_timewait_sock *tw;
	struct sock *sk;
	struct hlist_nulls_node *node;
	unsigned int slot;

	for (slot = 0; slot <= hashinfo->ehash_mask; slot++) {
		struct inet_ehash_bucket *head = &hashinfo->ehash[slot];
restart_rcu:
		cond_resched();
		rcu_read_lock();
restart:
		sk_nulls_for_each_rcu(sk, node, &head->chain) {
			if (sk->sk_state != TCP_TIME_WAIT)
				continue;
			tw = inet_twsk(sk);
			if ((tw->tw_family != family) ||
				atomic_read(&twsk_net(tw)->count))
				continue;

			if (unlikely(!atomic_inc_not_zero(&tw->tw_refcnt)))
				continue;

			if (unlikely((tw->tw_family != family) ||
				     atomic_read(&twsk_net(tw)->count))) {
				inet_twsk_put(tw);
				goto restart;
			}

			rcu_read_unlock();
			local_bh_disable();
			inet_twsk_deschedule(tw);
			local_bh_enable();
			inet_twsk_put(tw);
			goto restart_rcu;
		}
		/* If the nulls value we got at the end of this lookup is
		 * not the expected one, we must restart lookup.
		 * We probably met an item that was moved to another chain.
		 */
		if (get_nulls_value(node) != slot)
			goto restart;
		rcu_read_unlock();
	}
}
EXPORT_SYMBOL_GPL(inet_twsk_purge);
Commit	Line	Data
e48c414e ACM	1	/*
	2	* INET An implementation of the TCP/IP protocol suite for the LINUX
	3	* operating system. INET is implemented using the BSD Socket
	4	* interface as the means of communication with the user level.
	5	*
	6	* Generic TIME_WAIT sockets functions
	7	*
	8	* From code orinally in TCP
	9	*/
	10
172589cc	11	#include <linux/kernel.h>
9e337b0f	12	#include <linux/kmemcheck.h>
5a0e3ad6	13	#include <linux/slab.h>
3a9a231d	14	#include <linux/module.h>
e48c414e ACM	15	#include <net/inet_hashtables.h>
e48c414e ACM	16	#include <net/inet_timewait_sock.h>
696ab2d3	17	#include <net/ip.h>
e48c414e	18
13475a30	19
2a8875e7 ED	20	/**
	21	* inet_twsk_unhash - unhash a timewait socket from established hash
	22	* @tw: timewait socket
	23	*
	24	* unhash a timewait socket from established hash, if hashed.
	25	* ehash lock must be held by caller.
	26	* Returns 1 if caller should call inet_twsk_put() after lock release.
13475a30 ED	27	*/
	28	int inet_twsk_unhash(struct inet_timewait_sock *tw)
	29	{
	30	if (hlist_nulls_unhashed(&tw->tw_node))
	31	return 0;
	32
	33	hlist_nulls_del_rcu(&tw->tw_node);
	34	sk_nulls_node_init(&tw->tw_node);
2a8875e7 ED	35	/*
	36	* We cannot call inet_twsk_put() ourself under lock,
	37	* caller must call it for us.
	38	*/
13475a30 ED	39	return 1;
	40	}
	41
2a8875e7 ED	42	/**
	43	* inet_twsk_bind_unhash - unhash a timewait socket from bind hash
	44	* @tw: timewait socket
	45	* @hashinfo: hashinfo pointer
	46	*
	47	* unhash a timewait socket from bind hash, if hashed.
	48	* bind hash lock must be held by caller.
	49	* Returns 1 if caller should call inet_twsk_put() after lock release.
3cdaedae ED	50	*/
	51	int inet_twsk_bind_unhash(struct inet_timewait_sock *tw,
	52	struct inet_hashinfo *hashinfo)
	53	{
	54	struct inet_bind_bucket *tb = tw->tw_tb;
	55
	56	if (!tb)
	57	return 0;
	58
	59	__hlist_del(&tw->tw_bind_node);
	60	tw->tw_tb = NULL;
	61	inet_bind_bucket_destroy(hashinfo->bind_bucket_cachep, tb);
2a8875e7 ED	62	/*
	63	* We cannot call inet_twsk_put() ourself under lock,
	64	* caller must call it for us.
	65	*/
3cdaedae ED	66	return 1;
	67	}
	68
e48c414e	69	/* Must be called with locally disabled BHs. */
789f558c	70	static void inet_twsk_kill(struct inet_timewait_sock *tw)
e48c414e	71	{
789f558c	72	struct inet_hashinfo *hashinfo = tw->tw_dr->hashinfo;
e48c414e	73	struct inet_bind_hashbucket *bhead;
13475a30	74	int refcnt;
e48c414e	75	/* Unlink from established hashes. */
9db66bdc	76	spinlock_t *lock = inet_ehash_lockp(hashinfo, tw->tw_hash);
e48c414e	77
9db66bdc	78	spin_lock(lock);
13475a30	79	refcnt = inet_twsk_unhash(tw);
9db66bdc	80	spin_unlock(lock);
e48c414e ACM	81
e48c414e ACM	82	/* Disassociate with bind bucket. */
7f635ab7 PE	83	bhead = &hashinfo->bhash[inet_bhashfn(twsk_net(tw), tw->tw_num,
7f635ab7 PE	84	hashinfo->bhash_size)];
3cdaedae	85
e48c414e	86	spin_lock(&bhead->lock);
3cdaedae	87	refcnt += inet_twsk_bind_unhash(tw, hashinfo);
e48c414e	88	spin_unlock(&bhead->lock);
3cdaedae	89
05dbc7b5 ED	90	BUG_ON(refcnt >= atomic_read(&tw->tw_refcnt));
05dbc7b5 ED	91	atomic_sub(refcnt, &tw->tw_refcnt);
789f558c ED	92	atomic_dec(&tw->tw_dr->tw_count);
789f558c ED	93	inet_twsk_put(tw);
e48c414e ACM	94	}
e48c414e ACM	95
05dbc7b5	96	void inet_twsk_free(struct inet_timewait_sock *tw)
7054fb93	97	{
4dbc8ef7 ACM	98	struct module *owner = tw->tw_prot->owner;
4dbc8ef7 ACM	99	twsk_destructor((struct sock *)tw);
7054fb93	100	#ifdef SOCK_REFCNT_DEBUG
4dbc8ef7	101	pr_debug("%s timewait_sock %p released\n", tw->tw_prot->name, tw);
7054fb93	102	#endif
4dbc8ef7 ACM	103	kmem_cache_free(tw->tw_prot->twsk_prot->twsk_slab, tw);
	104	module_put(owner);
	105	}
	106
	107	void inet_twsk_put(struct inet_timewait_sock *tw)
	108	{
	109	if (atomic_dec_and_test(&tw->tw_refcnt))
	110	inet_twsk_free(tw);
7054fb93 PE	111	}
	112	EXPORT_SYMBOL_GPL(inet_twsk_put);
	113
05dbc7b5 ED	114	static void inet_twsk_add_node_rcu(struct inet_timewait_sock *tw,
	115	struct hlist_nulls_head *list)
	116	{
	117	hlist_nulls_add_head_rcu(&tw->tw_node, list);
	118	}
	119
	120	static void inet_twsk_add_bind_node(struct inet_timewait_sock *tw,
	121	struct hlist_head *list)
	122	{
	123	hlist_add_head(&tw->tw_bind_node, list);
	124	}
	125
e48c414e ACM	126	/*
	127	* Enter the time wait state. This is called with locally disabled BH.
	128	* Essentially we whip up a timewait bucket, copy the relevant info into it
	129	* from the SK, and mess with hash chains and list linkage.
	130	*/
	131	void __inet_twsk_hashdance(struct inet_timewait_sock tw, struct sock sk,
	132	struct inet_hashinfo *hashinfo)
	133	{
	134	const struct inet_sock *inet = inet_sk(sk);
463c84b9	135	const struct inet_connection_sock *icsk = inet_csk(sk);
81c3d547	136	struct inet_ehash_bucket *ehead = inet_ehash_bucket(hashinfo, sk->sk_hash);
9db66bdc	137	spinlock_t *lock = inet_ehash_lockp(hashinfo, sk->sk_hash);
e48c414e ACM	138	struct inet_bind_hashbucket *bhead;
	139	/* Step 1: Put TW into bind hash. Original socket stays there too.
	140	Note, that any socket with inet->num != 0 MUST be bound in
	141	binding cache, even if it is closed.
	142	*/
c720c7e8	143	bhead = &hashinfo->bhash[inet_bhashfn(twsk_net(tw), inet->inet_num,
7f635ab7	144	hashinfo->bhash_size)];
e48c414e	145	spin_lock(&bhead->lock);
463c84b9	146	tw->tw_tb = icsk->icsk_bind_hash;
547b792c	147	WARN_ON(!icsk->icsk_bind_hash);
e48c414e ACM	148	inet_twsk_add_bind_node(tw, &tw->tw_tb->owners);
	149	spin_unlock(&bhead->lock);
	150
9db66bdc	151	spin_lock(lock);
e48c414e	152
3ab5aee7	153	/*
05dbc7b5 ED	154	* Step 2: Hash TW into tcp ehash chain.
	155	* Notes :
	156	* - tw_refcnt is set to 3 because :
	157	* - We have one reference from bhash chain.
	158	* - We have one reference from ehash chain.
	159	* We can use atomic_set() because prior spin_lock()/spin_unlock()
	160	* committed into memory all tw fields.
3ab5aee7	161	*/
05dbc7b5 ED	162	atomic_set(&tw->tw_refcnt, 1 + 1 + 1);
05dbc7b5 ED	163	inet_twsk_add_node_rcu(tw, &ehead->chain);
3ab5aee7	164
05dbc7b5	165	/* Step 3: Remove SK from hash chain */
3ab5aee7 ED	166	if (__sk_nulls_del_node_init_rcu(sk))
3ab5aee7 ED	167	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
e48c414e	168
9db66bdc	169	spin_unlock(lock);
e48c414e	170	}
696ab2d3 ACM	171	EXPORT_SYMBOL_GPL(__inet_twsk_hashdance);
696ab2d3 ACM	172
216f8bb9	173	static void tw_timer_handler(unsigned long data)
c676270b	174	{
789f558c ED	175	struct inet_timewait_sock tw = (struct inet_timewait_sock )data;
	176
	177	if (tw->tw_kill)
	178	NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITKILLED);
	179	else
	180	NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITED);
	181	inet_twsk_kill(tw);
	182	}
	183
	184	struct inet_timewait_sock inet_twsk_alloc(const struct sock sk,
	185	struct inet_timewait_death_row *dr,
	186	const int state)
	187	{
	188	struct inet_timewait_sock *tw;
	189
	190	if (atomic_read(&dr->tw_count) >= dr->sysctl_max_tw_buckets)
	191	return NULL;
	192
	193	tw = kmem_cache_alloc(sk->sk_prot_creator->twsk_prot->twsk_slab,
	194	GFP_ATOMIC);
00db4124	195	if (tw) {
c676270b ACM	196	const struct inet_sock *inet = inet_sk(sk);
c676270b ACM	197
9e337b0f VN	198	kmemcheck_annotate_bitfield(tw, flags);
9e337b0f VN	199
789f558c	200	tw->tw_dr = dr;
c676270b	201	/* Give us an identity. */
c720c7e8 ED	202	tw->tw_daddr = inet->inet_daddr;
c720c7e8 ED	203	tw->tw_rcv_saddr = inet->inet_rcv_saddr;
c676270b	204	tw->tw_bound_dev_if = sk->sk_bound_dev_if;
66b13d99	205	tw->tw_tos = inet->tos;
c720c7e8	206	tw->tw_num = inet->inet_num;
c676270b ACM	207	tw->tw_state = TCP_TIME_WAIT;
c676270b ACM	208	tw->tw_substate = state;
c720c7e8 ED	209	tw->tw_sport = inet->inet_sport;
c720c7e8 ED	210	tw->tw_dport = inet->inet_dport;
c676270b ACM	211	tw->tw_family = sk->sk_family;
c676270b ACM	212	tw->tw_reuse = sk->sk_reuse;
81c3d547	213	tw->tw_hash = sk->sk_hash;
c676270b	214	tw->tw_ipv6only = 0;
f5715aea	215	tw->tw_transparent = inet->transparent;
c676270b	216	tw->tw_prot = sk->sk_prot_creator;
33cf7c90	217	atomic64_set(&tw->tw_cookie, atomic64_read(&sk->sk_cookie));
efd7ef1c	218	twsk_net_set(tw, sock_net(sk));
789f558c	219	setup_timer(&tw->tw_timer, tw_timer_handler, (unsigned long)tw);
47e1c323 ED	220	/*
	221	* Because we use RCU lookups, we should not set tw_refcnt
	222	* to a non null value before everything is setup for this
	223	* timewait socket.
	224	*/
	225	atomic_set(&tw->tw_refcnt, 0);
789f558c	226
eeb2b856	227	__module_get(tw->tw_prot->owner);
c676270b ACM	228	}
	229
	230	return tw;
	231	}
696ab2d3 ACM	232	EXPORT_SYMBOL_GPL(inet_twsk_alloc);
696ab2d3 ACM	233
696ab2d3 ACM	234	/* These are always called from BH context. See callers in
	235	* tcp_input.c to verify this.
	236	*/
	237
	238	/* This is for handling early-kills of TIME_WAIT sockets. */
789f558c	239	void inet_twsk_deschedule(struct inet_timewait_sock *tw)
696ab2d3	240	{
789f558c ED	241	if (del_timer_sync(&tw->tw_timer))
789f558c ED	242	inet_twsk_kill(tw);
696ab2d3	243	}
696ab2d3 ACM	244	EXPORT_SYMBOL(inet_twsk_deschedule);
696ab2d3 ACM	245
789f558c	246	void inet_twsk_schedule(struct inet_timewait_sock *tw, const int timeo)
696ab2d3	247	{
696ab2d3 ACM	248	/* timeout := RTO * 3.5
	249	*
	250	* 3.5 = 1+2+0.5 to wait for two retransmits.
	251	*
	252	* RATIONALE: if FIN arrived and we entered TIME-WAIT state,
	253	* our ACK acking that FIN can be lost. If N subsequent retransmitted
	254	* FINs (or previous seqments) are lost (probability of such event
	255	* is p^(N+1), where p is probability to lose single packet and
	256	* time to detect the loss is about RTO*(2^N - 1) with exponential
	257	* backoff). Normal timewait length is calculated so, that we
	258	* waited at least for one retransmitted FIN (maximal RTO is 120sec).
	259	* [ BTW Linux. following BSD, violates this requirement waiting
	260	* only for 60sec, we should wait at least for 240 secs.
	261	* Well, 240 consumes too much of resources 8)
	262	* ]
	263	* This interval is not reduced to catch old duplicate and
	264	* responces to our wandering segments living for two MSLs.
	265	* However, if we use PAWS to detect
	266	* old duplicates, we can reduce the interval to bounds required
	267	* by RTO, rather than MSL. So, if peer understands PAWS, we
	268	* kill tw bucket after 3.5*RTO (it is important that this number
	269	* is greater than TS tick!) and detect old duplicates with help
	270	* of PAWS.
	271	*/
696ab2d3	272
789f558c ED	273	tw->tw_kill = timeo <= 4*HZ;
789f558c ED	274	if (!mod_timer_pinned(&tw->tw_timer, jiffies + timeo)) {
696ab2d3	275	atomic_inc(&tw->tw_refcnt);
789f558c	276	atomic_inc(&tw->tw_dr->tw_count);
696ab2d3	277	}
696ab2d3	278	}
696ab2d3 ACM	279	EXPORT_SYMBOL_GPL(inet_twsk_schedule);
696ab2d3 ACM	280
b099ce26	281	void inet_twsk_purge(struct inet_hashinfo *hashinfo,
d315492b DL	282	struct inet_timewait_death_row *twdr, int family)
	283	{
	284	struct inet_timewait_sock *tw;
	285	struct sock *sk;
3ab5aee7	286	struct hlist_nulls_node *node;
575f4cd5	287	unsigned int slot;
d315492b	288
575f4cd5 EB	289	for (slot = 0; slot <= hashinfo->ehash_mask; slot++) {
	290	struct inet_ehash_bucket *head = &hashinfo->ehash[slot];
	291	restart_rcu:
738e6d30	292	cond_resched();
575f4cd5	293	rcu_read_lock();
d315492b	294	restart:
05dbc7b5 ED	295	sk_nulls_for_each_rcu(sk, node, &head->chain) {
	296	if (sk->sk_state != TCP_TIME_WAIT)
	297	continue;
d315492b	298	tw = inet_twsk(sk);
b099ce26 EB	299	if ((tw->tw_family != family) \|\|
b099ce26 EB	300	atomic_read(&twsk_net(tw)->count))
d315492b DL	301	continue;
d315492b DL	302
575f4cd5 EB	303	if (unlikely(!atomic_inc_not_zero(&tw->tw_refcnt)))
	304	continue;
	305
b099ce26 EB	306	if (unlikely((tw->tw_family != family) \|\|
b099ce26 EB	307	atomic_read(&twsk_net(tw)->count))) {
575f4cd5 EB	308	inet_twsk_put(tw);
	309	goto restart;
	310	}
	311
	312	rcu_read_unlock();
91035f0b	313	local_bh_disable();
789f558c	314	inet_twsk_deschedule(tw);
91035f0b	315	local_bh_enable();
d315492b	316	inet_twsk_put(tw);
575f4cd5	317	goto restart_rcu;
d315492b	318	}
575f4cd5 EB	319	/* If the nulls value we got at the end of this lookup is
	320	* not the expected one, we must restart lookup.
	321	* We probably met an item that was moved to another chain.
	322	*/
	323	if (get_nulls_value(node) != slot)
	324	goto restart;
	325	rcu_read_unlock();
d315492b	326	}
d315492b DL	327	}
d315492b DL	328	EXPORT_SYMBOL_GPL(inet_twsk_purge);